Germanium isotope fractionation during Ge adsorption on goethite ...

Available online at www.sciencedirect.com

ScienceDirect Geochimica et Cosmochimica Acta 131 (2014) 138–149 www.elsevier.com/locate/gca

Germanium isotope fractionation during Ge adsorption on goethite and its coprecipitation with Fe oxy(hydr)oxides Oleg S. Pokrovsky a,b,⇑, Albert Galy c, Jacques Schott a, Gleb S. Pokrovski a, Samia Mantoura c a

Geóscience and Environnement Toulouse, UMR 5563 CNRS, University of Toulouse, 14 Avenue Edouard Belin, 31400 Toulouse, France b BIO-GEO-CLIM Laboratory, Tomsk State University, Tomsk, Russia c Department of Earth Science, University of Cambridge, UK Received 27 July 2013; accepted in revised form 18 January 2014; Available online 1 February 2014

Abstract Isotopic fractionation of Ge was studied during Ge adsorption on goethite and its coprecipitation with amorphous Fe oxy(hydr)oxides. Regardless of the pH, surface concentration of adsorbed Ge or exposure time, the solution–solid enrichment factor for adsorption (D74/70Gesolution–solid) was 1.7 ± 0.1%. The value of the D74Gesolution–solid in Fe–Ge coprecipitates having molar ratio 0.1 < (Ge/Fe)solid < 0.5 remained constant at 2.0 ± 0.4%. For (Ge/Fe)solid ratio < 0.1, the D74Gesolution–solid increased with the decrease of Ge concentration in the solid phase, with the value as high as 4.4 ± 0.2% at (Ge/Fe)solid < 0.001, corresponding to the majority of natural settings. These results can be interpreted based on available structural data for adsorbed and coprecipitated Ge. It follows that Ge(OH)4° adsorption occurring as bidentate binuclear complexes at the goethite surface is characterised by an enrichment factor of 1.7%, likely related to the distortion of the GeO4 tetrahedron and the formation of Ge–O–Fe bonds at the goethite surface as compared to aqueous solution. In contrast, coprecipitation yields more distorted edge-sharing GeO4 tetrahedra and, in the case of the most diluted samples, part of the Ge is found in coordination 6, replacing Fe(III) in octahedral positions. This produces a greater enrichment of the solid phase in lighter isotopes, mostly due to the increase in Ge–O bond distances and coordination number compared to aqueous solution, which is in line with the basic principles of isotope fractionation. Discharge of hydrothermal fluids, leading to massive Fe(OH)3 precipitation in the vicinity of the springs should, therefore, represent an isotopically-heavy source of dissolved Ge to the ocean. Similarly, groundwater discharge and Fe(OH)3 precipitation at the Earth’s surface, Fe oxy(hydr)oxide formation in soils and riverine organo-ferric colloids coagulation, leading to iron hydroxide precipitation in estuaries, should produce an isotopically heavy Ge aqueous flux to the ocean compared to bedrock sources and particulate fluxes. Ó 2014 Elsevier Ltd. All rights reserved.

1. INTRODUCTION Because their outer electron structures are identical, germanium and silicon are known to exhibit similar ⇑ Corresponding author at: Geóscience and Environnement

Toulouse, UMR 5563 CNRS, University of Toulouse, 14 Avenue Edouard Belin, 31400 Toulouse, France. Tel.: +33 561332625; fax: +33 561332560. E-mail address: [email protected] (O.S. Pokrovsky). http://dx.doi.org/10.1016/j.gca.2014.01.023 0016-7037/Ó 2014 Elsevier Ltd. All rights reserved.

geochemical behaviours. As such, Ge has been widely used to trace both the continental and oceanic Si cycles (Froelich and Andreae, 1981; Froelich et al., 1985, 1992; Mortlock et al., 1993; Mortlock and Froelich, 1987; Filippelli et al., 2000; Derry et al., 2005, 2006). In this regard, Ge isotopes may provide useful geochemical tracers for the study of oceanic systems as well as continental weathering environments (Galy et al., 2003; Rouxel et al., 2006; Siebert et al., 2006; Qi et al., 2011; Escoube et al., 2012). Despite significant research efforts in deciphering Ge behaviour in

O.S. Pokrovsky et al. / Geochimica et Cosmochimica Acta 131 (2014) 138–149

natural systems such as hydrothermal systems (Mortlock et al., 1993; Evans and Derry, 2002; Wheat and McManus, 2008), iron-rich oceanic margin sediments (Kolodny and Halicz, 1988; King et al., 2000; Hammond et al., 2000; McManus et al., 2003), soils (Kurtz et al., 2002; Opfergelt et al., 2010; Cornelis et al., 2010), rivers (Murname and Stallard, 1990; Froelich et al., 1992; Derry et al., 2006) and biogenic minerals (Shemesh et al., 1989; Rouxel et al., 2006), the exact physical–chemical mechanisms controlling Ge at the Earth’s surface remain poorly characterized. One of the main results of previous studies of Ge isotope fractionation is that Ge is tightly linked to the Fe redox cycle and Fe mineral formation in contemporary and past Earth surface environments. As a result, the use of Ge/Si as a paleoceanographic tool remains uncertain as Ge may be removed from the oceans in iron-rich sediments independent of Si (King et al., 2000). Among major natural processes governing the fractionation of Ge and its isotopes, adsorption and coprecipitation onto/with Fe oxy(hydr)oxides are extremely important for the oceanic cycle of Ge as are the non-opaline Ge sink in marine sediments (Hammond et al., 2000; King et al., 2000; McManus et al., 2003), continental weathering and colloidal transport in natural waters. At pH representative of surface weathering conditions, Ge was shown to be preferentially sorbed to Fe-oxyhydroxide surfaces by a factor of 7 relative to Si (Anders et al., 2003). On the other hand, a study of Ge behaviour in soils along a climate gradient on Hawaii demonstrated that Ge sequestration is independent of Fe redox behaviour and the precipitation of Fe-oxyhydroxides is not a major factor contributing to Ge/Si fractionation during weathering compared to Ge partitioning into secondary clay minerals (Scribner et al., 2006). A recent mass balance of dissolved Si and Ge in the ocean showed that elementary fluxes are poorly constrained, but that the nonopaline Ge sink (likely to be adsorption and coprecipitation onto/with Fe oxy(hydr)oxides) could represent between 24% and 55% of the burial flux (Mantoura et al., submitted for publication). Therefore, experimental calibration of Ge isotope fractionation linked to its interaction with iron oxides/hydroxides will clearly provide further constraints on the Ge marine and continental budget and thus constitutes the first objective of this study. The sign and magnitude of the isotope fractionation of an element involved in a chemical reaction depend on the nature of the reaction (kinetic or equilibrium) and on the structures of the species and their associated energetic levels in the investigated reactants and products (Schauble, 2004). Structural characterization of Ge adsorbed at the goethite surface and coprecipitated with Fe oxy(hydr)oxide has been achieved using X-ray Absorption Spectroscopy (Pokrovsky et al., 2006). The present study extends this knowledge to Ge isotope fractionation during the adsorption and coprecipitation process. This should (1) allow identification and quantification of the physico-chemical factors controlling Ge isotopic variations in natural systems in the presence of Fe hydrous oxides, and (2) provide the first experimental constraints on Ge isotopic fractionation in the oxic aquatic environments of the Earth’s surface.

139

2. MATERIALS AND METHODS 2.1. Adsorption and coprecipitation Goethite powder, synthesized following a procedure described by Cornell and Schwertmann (1996), had a specific surface area of 23.2 m2/g (see Pokrovsky et al., 2006 for its detailed properties). All solutions were prepared from ultrapure NaNO3, HNO3, NaOH and MilliQ water. Germanium stock solution was prepared by dissolving germanium dioxide (Fluka puriss. for electronic purposes, 99.99%) in deionized MilliQ water. Batch adsorption experiments were performed in 1 mM NaNO3 background electrolyte solution using acid-cleaned 30 mL polypropylene (PP) vials, which were agitated for 2 days on a RotaMagÒ mixer at 25 ± 0.5 °C. Four concentrations of goethite powder were used: 1.8, 3.3, 6.6 and 17 g/L, and initial Ge concentration in solution ranged from 344 to 384 lM. At the end of experiments, the solution pH was measured, the suspension was centrifuged for 15 min at 2500g, filtered through pre-cleaned 0.22 lm acetate cellulose sterile disposable filters, and the supernatant was acidified by bi-distilled HNO3 prior to analyses. During filtration, the first 5–10 mL of the filtrate were discarded to allow filter material rinsing and porespace saturation. In order to evaluate possible artifacts linked to Ge chemical and isotopic fractionation during sample filtration, standard Ge solution of neutral pH and concentration similar to that in experimental samples, was filtered and analysed for total Ge concentration and Ge isotope ratio (sample “Blank Ge no ads”). This filtration procedure did not modify dissolved Ge concentration more than 2%, and the isotopic ratio of filtered standard solution was similar, within ± 0.1%, to d74Ge of the unfiltered initial solution. The kinetics of Ge adsorption was studied in a batch stirred reactor through which N2 was continuously bubbled. N2 bubbling was intended to avoid oxygen interferences with goethite surface, but mostly to remove the dissolved CO2 that is capable of forming adsorbed carbonate and bicarbonate complexes, and thus compete with Ge for goethite binding sites. In these experiments, 1 mM NaNO3 solution with 6.6 g/L of goethite powder was used, and pH was kept constant at 8.50 ± 0.05. Aliquots of goethite suspension were removed and immediately filtered for subsequent chemical and isotopic analyses. Germanium coprecipitation with iron oxy(hydr)oxide was performed at 25 °C in 30 mL PP vials via slow oxidation of the 0.8 mM Fe(II) sulfate solutions in 0.1 M NaNO3 by atmospheric oxygen. Values of pH varied from 4.6 to 10.4, and the initial Ge concentration ([Ge]o) varied from 16 to 930 lM. Amorphous iron oxy(hydr)oxide was formed during the first 0.5–1.0 h of reaction and allowed to age for 24 h. This yielded Ge concentration in the solid phase corresponding to molar ratio Ge/Fe of 0.02–0.5. Note that aging of precipitates for 1–14 days at 25 °C in contact with solutions open to the atmosphere did not produce any significant change in Ge and Fe aqueous concentrations. Another method of Ge coprecipitation with iron hydroxide consisted of neutralizing a 900 lM Fe(III) acidic solution (made from Fe(NO3)3 in 0.1 M NaNO3 at pH 2) by the


addition of 2% NH4OH until pH reached 6.5–7.5. All precipitates were separated from solution by centrifugation for 10 min at 4000 rpm, rinsed in MilliQ water, and dried for 72 h at 60 °C. 2.2. Chemical analyses Solution pH was measured using a combination glass electrode (Mettler Toledo) calibrated on the activity scale with NIST pH buffers (pH = 4.006, 6.865, and 9.180 at 25 °C). The precision of pH measurements was ±0.002 units (0.1 mV). Total germanium ([Ge]tot) concentration was measured by flame atomic absorption using a Perkin Elmer 5100 PC spectrometer in the concentration range of 30–800 lmol L1 with an uncertainty of 2%. For Ge analyses in the concentration range of 1.4–140 lmol L1, a colorimetric method with molybdenum blue was used (uncertainty of 2%). Finally, at very low concentrations ([Ge] < 7 lmol L1), dissolved germanium was determined by ICP-MS (Elan 6000, Perkin Elmer) using 70Ge and 74 Ge isotopes. Indium and rhenium were used as internal standards, and the international geostandard SLRS-4 (Riverine Water Reference Material for Trace Metals certified by the National Research Council of Canada) was used to check the validity and reproducibility of the analyses. This freshwater standard has low concentrations of major ions and was similar in ionic strength to the majority of our experimental solutions. The uncertainty of ICPMS analyses was 10% at [Ge] < 0.014 lmol L1and 5% at [Ge] P 0.014 lmol L1. The three methods of Ge analysis employed in this study agreed within 5%. The Ge blank in our experiments, a typical aqueous solution prepared as in the experiments but without Ge added, was between 30 and 60 pg. Total dissolved iron (Fe(II) + Fe(III)) was measured by flame atomic absorption spectroscopy (AAS) using a Perkin Elmer 5100 PC spectrometer in the concentration range of 3.6–90 lmol L1 with an uncertainty of 2% and a detection limit of 0.9 lmol L1. Fe concentration in the solid was measured by AAS after dry sample dissolution in concentrated HNO3, with an uncertainty of 5%. This uncertainty corresponded to the mean 2r of five repeat measurements of an in-house standard solution.

n o dx Geð&Þ ¼ ðx Ge=70 GeÞsample =ðx Ge=70 GeÞStandard 1 1000

where x = 72, 73 or 74. Because of the lack of an international isotopic standard of germanium, a commercial germanium standard solution (with a germanium concentration of 1000 lg g1) in 2% KOH (Aldrich, USA) was used as our in-house isotopic standard. Its long-term reproducibility was close to 0.1% for d74Ge. Filtered aqueous solutions of adsorption experiments (Fig. 1) were measured in dry plasma mode, where samples and standards were introduced into the plasma torch through an Aridus desolvating nebuliser (Galy et al., 2003). This device minimises the introduction of H2O, CO2, O2 and N2 into the plasma. Elution on a column of AG 50 W-X8 cationic resin with 0.5 M HNO3 allowed the separation of germanium from the alkali matrix. The chemistry yield was found to be P90% as quantified by comparing the ICP-MS measured total dissolved Ge concentration in solution and converting in total mass of Ge before and after the ion exchange column separation procedure. The external repeatability (replicate analyses on different days) of in-house isotopic standard (commercial germanium standard solution) measured over 8 months was observed to be 0.12%, 0.16% and 0.07% for d74Ge d73Ge and d72Ge,

A

2.5

[Ge]ads , µmol/m²

140

2 1.5 1 isotopic analysis

0.5 0 0.1

1

10

100

1000

10000

Elapsed time, min

B

100

Ge-isotope ratios were measured using a Nu Instruments MC-ICPMS and a sample-standard bracketing protocol (Galy et al., 2003). Briefly, 70Ge, 72Ge, 73Ge and 74 Ge were collected in Faradays Low 3, Axial, High 2 and High 4 respectively within the overall array. Standard and sample isotope values were measured 4 and 3 times, respectively for 200 s each. Each 200 s measurement consisted of 5 s integration of the electronic background by deflecting the ion beam at the electrostatic analyser followed by 100 s acquisition of the signal, repeated twice. This protocol allows the calculation of 6 brackets, and each value corresponds to their average. Isotopic compositions are expressed as a permil deviation from the isotopic composition of the standard solution as follows:

[Ge]ads, %

80

2.3. Ge isotopic measurements

60

17 g/L

6.6 g/L

3.3 g/L

1.8 g/L

40 20 0 2

4

6

8

10

12

pH Fig. 1. Concentration of adsorbed germanium in contact with goethite as a function of time, obtained from aqueous solution analysis (A) and pH dependence of Ge adsorption on goethite at various solid/solution ratios (B). In both cases, the final solution of a separate experiment was analysed. The error bars are within the symbol size. Aqueous solution samples used for isotopic analysis are circled.


respectively. The standard-sample-standard bracketing technique used to examine repeatability of measurements yielded a stability of the uncorrected 74Ge/70Ge ratio during extended runs of up to 12 h of close to 0.12% per hour. As a part of the evolution of the Ge isotopic method, selected coprecipitation experiments (Fig. 2) were measured using a continuous flow hydride generation system coupled to the MC-ICPMS (Rouxel et al., 2006, Mantoura et al., submitted for publication). The reducing agent (a solution of 0.26 N of NaBH4 stabilized in 0.0125 N of NaOH) and the sample were introduced to the hydride generation (HG) system by a peristaltic pump. The separation of the gas from the liquid has been achieved with the use of a modified Scott-type spray chamber cooled at 4 °C and a PTFE filter between the spray chamber and the ICP-MS torch. Stability of hydride formation was improved by the use of a mixing coil (20 cm), consistent pumping of the liquid waste to the drain, and a second Ar inlet placed between the spray chamber and the ICP-MS torch. The decontamination of the inlet was achieved by the use of a solution of diluted HNO3 (2%) for 5 min. Given that all the solutions introduced into MC ICP MS were prepared at similar Ge concentration and that the relative isotopic difference between different samples was within few%, the cross-contamination of the succeeding sample was 0.05. For 0.02 6 (Ge/Fe)solid 6 0.05, well crystallized lepidocrocite was identified in our samples, that

Table 1 Synthesis of all conducted experimental and the range of experimental conditions. Experiment no.

Type of interaction

Duration

Goethite (g/L) or [Fe]aq (mg/L)

pH range

Ge range, lM

1 Ads to 7 Ads IV-1 to IV-5 5–1 to 5–17 4–2 to 4–16 5–3 to 5–18 C-16 C-17

Adsorption Adsorption Adsorption Coprecipitation during Fe(II) oxidation Coprecipitation during Fe(II) oxidation Coprecipit., Fe(III) Hydrolysis Coprecipit., Fe(II) oxidation

48 h 48 h 0.1–150 h 26 h 24 h 26 26

3.3–17 g/L 1.8 g/L 6.6 g/L 13–45 mg/L 0.01–160 mg/L 10.5 mg/L 93 mg/L

2.9–11.3 2.7–10.7 8.5 5.0–5.3 5.0–10.4 6.7 4.6

24–207 250–332 39–73 16–930 0.1–160 52–275 200–275

Table 2 Experimental data measured in this study. The 2r standard deviation is based on 3–4 measurements of the same sample. Adsorption

% Geads

d74Ge & 2r

73

39 85 24 82 207 207 194

2.9 5.5 8.16 11.31 2.5 7.18 9.20

17 6.6 6.6 6.6 6.6 3.3 3.3

0.8 1.7 2.1 1.7 0.9 1.8 2.0

48 48 48 48 48 48 48

0.01 1.40 1.09 1.56 1.06 0.72 0.71 0.74

0.09 0.02 0.02 1.01 0.02 0.80 0.10 1.17 0.32 0.78 0.16 0.54 0.04 0.52 0.01 0.55

0.08 0.01 0.05 0.68 0.03 0.55 0.06 0.78 0.21 0.53 0.09 0.35 0.05 0.34 0.05 0.38

0.06 9 0.03 2 0.04 10 0.07 3 0.19 3 0.07 6 0.03 3 0.02 3

0.19 0.37 0.13 0.34 1.10 1.08 0.97

1.58 ± 0.07 1.46 ± 0.10 1.69 ± 0.09 1.39 ± 0.35 1.82 ± 0.40 1.79 ± 0.20 1.72 ± 0.25

IV-0 Initial soln. 0 IV1 13.0 IV-2 20.4 IV-3 27.3 IV-4 30.0 IV-5 28.2

344 381 384 370 357 376

332 306 269 250 270

2.68 3.98 6.70 8.97 10.7

1.8 1.8 1.8 1.8 1.8

1.2 1.9 2.4 2.6 2.5

48 48 48 48 48

0.10 0.09 0.22 0.40 0.51 0.38

0.07 0.09 0.05 0.07 0.05 0.14 0.07 0.30 0.10 0.34 0.10 0.29

0.05 0.06 0.05 0.03 0.04 0.10 0.09 0.19 0.06 0.25 0.11 0.19

0.04 0.04 0.03 0.03 0.06 0.07

6 3 3 3 4 5

1.43 1.38 1.46 1.53 1.35

1.52 ± 0.40 1.60 ± 0.37 1.86 ± 0.25 2.04 ± 0.30 1.73 ± 0.35

Exp no. 5, Kinetics in 0.001M NaNO3 5–1 78.8 5–2 81.0 5–6 84.3 5–11 85.7 5–17 88.6

343 343 343 343 343

73 65 54 49 39

8.5 8.5 8.5 8.5 8.5

6.6 6.6 6.6 6.6 6.6

1.8 1.8 1.9 1.9 2.0

0.1 0.2 2.8 23 150

0.11 0.09 0.11 0.09 0.05

0.06 0.10 0.20 0.14 0.04

0.04 0.05 0.08 0.04 0.03

2 2 3 3 2

0.37 0.36 0.28 0.22 0.19

1.71 ± 0.13 1.83 ± 0.11 1.76 ± 0.12 1.55 ± 0.11 1.67 ± 0.07

[Ge]final pH [Ge]init lmol L1 lmol L1

(Ge/Fe) mol

Fe, mg/L d74Ge&

16.1 456 927 160 160 160 160 275 275

0.030 0.292 0.493 0.097 0.134 0.141 0.178 0.276 0.047

Co-precipitation Blank 4–5 4–2b 4–12 4–16 5–3 5–7 5–9 5–18 C-16a C-17b a b

Initial solution Fe(II)SO4 oxidation At different Ge0 [Fe]init = 45 mg/L Fe(II)SO4 oxidation [Fe]init = 45 mg/L Fe(III) hydrolysis Fe(II) oxidation

322 649 119 87 47 17 52 196

4.99 5.14 5.27 4.97 5.17 5.59 10.38 6.68 4.59

45 14.5 19.2 13.2 21.2 14.4 2.52 0.01 10.5 93

0.05 1.62 0.77 0.54 0.76 1.17 1.56 1.28 1.74 1.11

2r

1.00 1.09 1.07 0.97 1.16 d73Ge&

0.01 0.02 0.77 1.27 0.32 0.55 0.07 0.41 0.25 0.55 0.30 0.88 0.10 1.19 0.37 0.96 0.18 1.32 0.06 0.85

2r 0.00 0.58 0.19 0.07 0.14 0.22 0.14 0.29 0.14 0.04

0.65 0.72 0.73 0.63 0.74 d72Ge & 0.04 0.85 0.37 0.27 0.39 0.59 0.80 0.65 0.91 0.59

2r 0.07 0.36 0.16 0.05 0.11 0.13 0.09 0.17 0.10 0.04

Nb d74Gesolid& 2 2 4 3 3 3 3 3 3 2

0.05 1.67 1.08 1.98 1.29 0.57 0.10 0.34 2.53

D74Ge solution–solid 1.57 ± 0.80 2.44 ± 0.60 1.62 ± 0.26 2.74 ± 0.75 2.46 ± 0.66 2.14 ± 0.14 1.38 ± 0.41 2.08 ± 0.52 3.64 ± 0.20


344 344 344 344 344 344 344 344

1.34 1.47 1.48 1.33 1.47

Nb d

Gesolid & D74Ge solution–solid

Blank Ge no ads 0 1 Ads 88.7 2 Ads 75.2 3 Ads 92.9 4 Ads 76.3 5 Ads 40.0 6 Ads 40.0 7 Ads 43.7

d

Ge & 2

74

Goethite Geads Time, h g/L lmol/m2

d

Ge & 2r

72

[Ge]init [Ge]final pH lmol L1 lmol L1

Fe(III) hydrolysis (1 mM Fe3+). Fe(II) oxidation, [Fe2+]init = 188 mg/L.

143

144

O.S. Pokrovsky et al. / Geochimica et Cosmochimica Acta 131 (2014) 138–149 1.8 g/L

A 1.8

3.3 g/L

1.6

6.6 g/L

74

δ Gesolution , ‰

1.4

6.6 g/L, kinetics 0.1-150 hrs

1.2

17 g/L

1.0 0.8 0.6 0.4 0.2 0.0 0

20

40

60

80

100

% adsorbed Ge

B

Fig. 4. Isotopic shift of 74Ge between solution and solid measured in adsorption experiments as a function of Ge surface concentration (A) and solution pH (B). Fig. 3. Plot of the aqueous d74Ge value (A) and isotopic shift (B) between aqueous solution and solid phase for Ge adsorption at the surface of goethite. Diamonds, squares, triangles, open and closed circles denote experiments with different goethite concentration of 1.8, 3.3, 6.6 (reversible adsorption and kinetics) and 17 g/L, respectively.

Overall, the D74Gesolution–solid characteristic of coprecipitation processes was 0.3–2% higher than that associated with Ge adsorption on goethite. 4. DISCUSSION 4.1. Factors responsible for Ge isotope fractionation

recrystallized into goethite upon aging in mother solution at ambient temperature (Pokrovsky et al., 2006). The value of d74Ge in the fluid phase is positively correlated (r2 = 0.68) with the amount of Ge coprecipitated with iron hydroxide (Fig. 5A). The D74Gesolution–solid is not correlated to the pH of the solution (Fig. 5B) or the (Ge/Fe)solid ratio (Fig. 6) at 95% confidence (r2 < 0.5). At the same time, it is not excluded that the most and the least fractionated samples are retrieved from the lowest and the highest pH experiments, respectively (Fig. 5B). There are two trends of D74Gesolution–solid dependence on (Ge/Fe)solid: a constant value at (Ge/Fe)solid > 0.2 and a linear (r2 = 0.97 at p = 0.95) trend at (Ge/Fe)solid < 0.2 with an intercept at +4.43%. A distinct difference in D74Gesolution–solid is observed for the most diluted samples (3.19 ± 0.48% for Ge/ Fe < 0.1, calculated as the average of two low concentration samples C-17 and 5–3, Table 2), compared to the most concentrated ones (2.02 ± 0.43% for 0.1 < Ge/Fe < 0.5). The precipitation mode, either Fe(II) oxidation or Fe(III) hydrolysis, does not have any significant effect on the resulting D74Gesolution–solid value (Figs. 5 and 6). The effect of the crystallinity of the formed precipitate, lepidocrocite versus amorphous Fe hydroxide, on Ge isotopic fractionation between aqueous solution and resulting precipitates could not be resolved beyond the uncertainty of our measurements.

Several structural/chemical factors may be responsible for Ge isotope fractionation during adsorption and coprecipitation processes in aqueous solution: (i) germanic acid protonation in solution and at the mineral surface; (ii) goethite surface coverage, which may control Ge binding to weak sites of high abundance versus low abundance strong sites; (iii) Ge binding mode, i.e., bidentate versus monodentate and binuclear versus mononuclear; (iv) Ge–O bond distances, the identity of the second neighbor, and the degree of distortion of the 1st coordination sphere; and finally, (v) Ge coordination in solution and in the solid phase, i.e., tetrahedral versus octahedral. All or some of these factors determine the equilibrium isotope fractionation between Ge at the goethite surface or in the Fe oxy(hydroxide) lattice and aqueous solution and control the fractionation kinetics during irreversible coprecipitation phenomena. The effect of the first factor, germanic acid protonation/ deprotonation at the goethite surface on Ge isotopic fractionation can be tested by comparing the D74Gesolution–solid among experiments of different equilibrium pH (Fig. 4B). According to the surface complexation model for Ge speciation at the goethite surface (Pokrovsky et al., 2006), P90% of adsorbed Ge is present as neutral > FeO– Ge(OH)3° species at pH < 5–6 whereas in alkaline solutions,


Fig. 5. Plots of the isotopic composition of the fluid phase as a function of Ge fraction coprecipitated in the solid (A) and the shift between solution and solid phase versus solution pH (B). The horizontal error bars are within the symbol size (from 1% to 2% in A and 0.01 pH units in B) unless shown.

Fig. 6. Effect of the Ge/Fe ratio in the solid phase on overall 74Ge isotopic shift between aqueous solution and precipitated Fe oxy(hydr)oxide. The vertical error bars were numerically calculated by error propagation whereas the horizontal error bars mostly reflect the analytical uncertainty. Arrowed dashed line shows a linear extrapolation, without taking into account sample 4–2. The left-hand end of this dashed line corresponds to the low concentrations of natural samples, which yields d74Gesolution–solid = 4.43 ± 0.20& at Ge/Fe < 0.001.

at pH of 10, half of the adsorbed Ge is represented by deprotonated > FeO–Ge(OH)2 species. Because the solution pH exhibits a minor effect on d74Gesolution–solid (Fig. 5B), it can be concluded that the degree of

145

protonation of Ge(O,OH)4 tetrahedra at the goethite surface does not impact Ge isotopic fractionation between aqueous solution and the goethite surface. The role of the 2nd factor is also likely to be small as follows from the lack of a statistically significant relationship between D74Gesolution–solid and the Ge surface concentration (30–100% of goethite surface site occupied, Fig. 4A). This strongly suggests that the degree of goethite surface coverage by sorbed Ge that can encounter either weak or strong binding sites does not affect the D74Gesolution–solid value for adsorption. The 3rd and 4th factors can be tested using available data on the structure of the complexes formed on goethite surface by germanic acid and other similar neutral molecules and oxyanions (Pokrovsky et al., 2006 and references therein). These data suggest that Ge is adsorbed in the form of binuclear complexes, and 1:1 stoichiometry between Ge(aq) and >FeOH surface sites is always maintained as two Fe centers interact with two Ge tetrahedra. This model is supported by XANES and EXAFS results demonstrating (1) the tetrahedral Ge environment (within ±10% of XAS uncertainty) for all adsorption samples, and (2) the distance ˚ between Ge and 0.5–1 Fe atom as the second of 3.3 ± 0.1 A neighbor corresponding to double-corner bi-dentate (C2-type) complexes between the Ge-tetrahedra and two adjacent Fe-octahedra sharing a common edge at the goethite surface (Fig. 7A). At the same time, given the average nature of an EXAFS signal, Ge atomic environment may be multiple, particularly at the high surface coverage here, with different complexes simultaneously present at the FeOOH surface as shown in Fig. 7B. The enrichment of lighter isotopes at the surface during the adsorption reaction can be explained by the change of next-nearest environment of the adsorbed atom. Indeed, even at similar Ge–O distances between aqueous and adsorbed Ge tetrahedra, the appearance of the second neighbour, such as Fe in adsorbed samples, leads to a decrease in the symmetry and an increase in the disorder compared to the aqueous Ge tetrahedron. Second, the formation of bidendendate tetrahedral binuclear complexes (C2-type, Fig. 7), and tetrahedral complexes sharing their edges with Fe octahedra (E-type, Fig. 7), is likely to increase their distortion and the disorder among Ge positions due to higher number of possible geometries. This should be accompanied by a decrease of the stability of Ge-ligand bonds in the first coordination sphere and, in accord with the principles of quantum mechanics, by an enrichment of the solid phase in lighter isotopes. Similar mechanisms of solid phase enrichment in lighter isotopes compared to the aqueous solution could operate for other oxyanions and neutral molecules such as MoO42 (Barling and Anbar, 2004) and Si(OH)4 (Delstanche et al., 2009). For example, Mo adsorbed on birnessite is known to form a polynuclear structure with distorted octahedral coordination compared to tetrahedrally coordinated MoO42 in aqueous solution, which produces 2.7% in d98/95Mo enrichment in lighter isotopes at the surface (Wasylenki et al., 2011). The effect of the last factor, the tetrahedral (GeIV) versus octahedral (GeVI) environment, on Ge isotopic fractionation during its interaction with Fe oxy(hydr)oxide could

146


Fig. 7. Schematic structures of the atomic environment of Ge(IV) adsorbed onto or co-precipitated with Fe(III) oxy-hydroxides. (A): Bidentate binuclear adsorbed complexes that maintain 1:1 stoichiometry between Fe surface centers and Ge(OH)4 in case of adsorption experiments, notably at high surface coverages, producing average fractionation D74Gesolution–solid of 1.69 ± 0.08&. (B): Bidentate mononuclear (E-type) and bidentate binuclear (C2-type) complexes of tetrahedral Ge and octahedral [6]Ge substitution for Fe octahedral in the bulk of Fe oxy(hydr)oxides in case of Ge coprecipitation with Fe oxy(hydr)oxide, producing possible fractionation D74Gesolution–solid of 4.43 ± 0.20&. Lightgrey octahedra denote the Fe(O/OH)6 coordination sphere with the metal in the centre; blue tetrahedra and yellow octahedra stand for Ge(IV) four- and six-coordinated with O/OH, respectively. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

only be probed on coprecipitated samples that exhibited a partial GeVI environment (Pokrovsky et al., 2006). The significant increase of D74Gesolution–solid for samples C-17 and 5–3 (Table 2), which correspond the lowest Ge concentration in the solid phase, may be linked to the change of Ge local structure in these diluted samples. Indeed, according to XAS results, samples with Ge/Fe < 0.1 exhibit a splitting of the first atomic shell, with Ge in both tetrahe˚ , N = 3.0–3.5) and octahedral dral (R = 1.77 ± 0.02 A ˚ (R = 1.92 ± 0.03 A, N = 0.5–1.3) coordination with oxygen, in contrast to the samples having a Ge/Fe > 0.1 (Pokrovsky et al., 2006). It is therefore likely that esolution–solid will tend to be greater for GeO6, which exhibits larger Ge–O distances and thus weaker Ge–O bonds compared to GeO4 both in the aqueous solution and the solid (i.e., see Criss, 1999; Schauble et al., 2004). If we assume that

esolution–solid for GeVI (esolution–solidVI) is greater than esolution–solid for GeIV (esolution–solidIV) but remained constant through the coprecipitation and that the experiments are at isotopic equilibrium, then it can easily be demonstrated that the observed d74Gesolution–solid is linearly correlated with the proportion of GeVI in Ge. Given that the fraction of octahedral Ge in samples studied in this work is (1) rather low, and (2) only observed in solid with Ge/Fe < 0.1 suggesting a very low GeVI/Fe, the Ge/Fe ratio of the solid is a good approximation of the proportion of GeVI in the solid. As a result, the extrapolation of d74Gesolution–solid to (Ge/ Fe)solid = 0 will give a good approximation of esolution–solidVI (Fig. 6). This extrapolation was performed for the most diluted samples with (Ge/Fe)solid < 0.2 and yielded d74Gesolution–solid = 4.4 ± 0.2% (2r at p = 0.95) in infinitely diluted sample (r2 = 0.97 without sample No. 4–2). This value should be regarded as tentative given (i) the low number of experimental data points used to carry out the correlation and (ii) the high experimental uncertainties on individual data points used to calculate the extrapolated value. Overall, the obtained experimental results confirm recent first-principles density functional theory calculation of Ge adsorbed onto Fe oxy(hydr)oxide surfaces (Li and Liu, 2010). These authors demonstrated that Ge isotopes can be distinctly fractionated by such adsorption processes, to about 1.7%. It is interesting to note that the esolution–solid for Si, which forms bidentate binuclear complexes at the goethite surface similar to germanium complexes (Vempati and Loeppert, 1989; Pokrovsky et al., 2006) is 2–3 times lower (D29Sisolution–solid = 0.54 and 0.81%; Delstanche et al., 2009) than the corresponding esolution–solid for Ge, whereas the differences in the reciprocal of the square root of isotope masses for 74/70Ge and 29/28Si (Criss, 1999 and reference therein) are almost the same. The lighter isotopic composition of adsorbed Ge compared to Si may be explained by the more distorted structure of the Ge complexes present at the goethite surface. Indeed, the formation of bidendate binuclear (C2-type) Si and Ge complexes yields a distance between two oxygen atoms at the goethite surface equals ˚ (Szytula et al., 1968; Manceau, 1995) and Si–O to 2.6 A ˚ (Newton and and Ge–O bond lengths equal to 1.65 A ˚ Gibbs, 1980) and 1.76 A (Pokrovsky et al., 2006), respectively. This induces a decrease of the O–(Si,Ge)–O angle from 109.5° in the aqueous species to 104° and 95° in the adsorbed Si and Ge tetrahedral, respectively. At the same time, the Mo oxyanion is also known to form complexes similar to Ge (Bibak and Borggaard, 1994), and exhibits comparable esolid-solution (2%, Barling et al., 2001; Siebert et al., 2003; Barling and Anbar, 2004; Goldberg et al., 2009; Wasylenki et al., 2011) despite a difference in the reduced mass (smaller by 68%). An important and poorly studied aspect of Ge geochemistry in natural ferric oxy(hydr)oxides is the presence in the latter of Si and/or Al as impurities (e.g., Blanch et al., 2008; Silva et al., 2010). These trace components may certainly compete with Ge for specific binding sites at the goethite surface and thus change the overall fractionation factor between the mineral and the aqueous solution, as is known for other oxyanions (Silva et al., 2010). However, we do not


anticipate a significant effect of these impurities as our adsorption data do not suggest the existence of strong (low abundant) versus weak (highly abundant) binding sites for Ge isotopes (Figs. 3B and 4A). As such, any competition between Ge and these components in nature is rather unlikely. 4.2. Geochemical applications Results obtained in the present study allow straightforward evaluation of the degree of Ge isotope fractionation in natural settings where the formation of Fe oxy(hydr)oxide occurs. Two basic cases can be considered, when Ge-bearing waters interact with Fe-rich sediments or soils and when co-precipitation of Ge occurs during the oxidation of Fe(II) and the massive formation of Fe(III)oxyhydroxide. The majority of natural settings can be approximated by the room temperature used in the present study to quantify Ge isotope fractionation between aqueous solution and Fe oxyhydroxides. Since diffusion (transport) processes do not control Ge adsorption and coprecipitation, as follows from fast reaction kinetics (Pokrovsky et al., 2006), we do not expect a significant effect of temperature on isotope fractionation. However, similar to other oxyanions (Wasylenki et al., 2008), some decrease of the isotopic fractionation with temperature increase may occur. Note that the first natural case, the infiltration of Gebearing fluids through Fe(III)-bearing soils or sediments, should prompt an enrichment of the fluid phase in heavier isotopes, proportional to the amount of Ge, up to the esolution–solid value of 1.69 ± 0.08%. Such a Ge isotope fractionation can be expected in groundwaters in contact with Fe(III)-rich clays or sandstones or, in the case of surface water interaction, with saprolite soils. Thus, tropical rivers draining highly weathered magmatic rocks could be enriched in d74Ge by 1–1.5% relative to the unaltered bedrocks. In the case of Fe oxy(hydr)oxide precipitation in marine or freshwater sediments within pore fluids or bottom water layers, the enrichment of the liquid phase in heavier isotopes should be much higher given the possibility of repeated cycles and Rayleigh distillation in a closed pore space, as is known for other isotopic couples; e.g., d57Fe values can vary up to 7% in sediment porewaters of subterranean estuaries (Rouxel et al., 2008). Ge coprecipitation with Fe oxyhydroxide includes the following settings (i) hydrothermal spring discharges in the ocean or on the continent, within the volcanic provenances, where massive Fe oxy(hydr)oxide formation occurs mediated by Fe(II) oxidation; (ii) secondary Fe hydroxide formation in soils; (iii) acid mine drainage and Fe(III) precipitation along the flow path; (iv) Fe(II) oxidation during groundwater discharge at the earth surface; and (v) Fe(III) coagulation in the estuarine zone. The first example is found at the top of the plumes above hydrothermal vents near oceanic ridges (i.e., Mortlock et al., 1993) where significant amounts of Ge can be retained by iron hydroxide, thus enriching the output fluid and oceanic water in heavier isotopes (Mantoura et al., submitted for publication). Both acid mine drainage and groundwater Fe(II) oxidation lead

147

to preferential coprecipitation of lighter Ge isotopes with newly formed amorphous Fe oxy(hydr)oxides, thus enriching the small rivers, seepage zones, and streams in heavier isotopes. Indeed, the emergence of reduced underground waters is often followed by the precipitation of iron oxy(hydr)oxides in the vicinity of springs. In such settings, a decrease of the Ge/Si ratio in the fluid induced by the preferential scavenging of Ge has already been invoked (Anders et al., 2003). Moreover, Ge-rich goethite and hematite from the weathering zone of the copper-arsenic sulfide Apex Mine (Utah, USA) were reported to contain octahedrally coordinated Ge in samples with (Ge/Fe)solid 0.01 (Bernstein and Waychunas, 1987). Our experimental results strongly suggest that in such a setting, secondary minerals will have d74Ge ca. 3–4% lower than the d74Ge of the fluid and, presumably, the source rock. The process of oxidation of groundwater Fe(II) is also known to occur at the riparian or the hyporheic zones of organic-rich streams in the boreal zone, where formation of Fe(III)-colloids stabilized by dissolved organic matter of allochthonous origin occurs (Pokrovsky and Schott, 2002). These colloids (1 kDa–0.22 lm in size), being represented by Fe(III) hydroxide globules with incorporated trace elements, are likely to have light Ge isotopic signatures relative to the truly dissolved low molecular weight (